Research

Research Projects

Smart parking and vehicle tracking project
- Propose AI model that performs dehazing, low-light enhancement, and superpixel on livestream CCTV
- Introduce automatic license plate recognition model with the recognition rate of over 95%
- Research and develop a software that supports vehicle tracking and smart parking.
Tunnels defect detection based on deep learning and explainable AI technology
- Train a deep learning model to detect and measure cracks/defects
- Explain, visualize, and interpret the deicisions made by the proposed model
Context-adaptive neural network based prediction for HEVC image compression
- Propose a set of neural network models for intra image prediction
- These models are trained on a large unconstrained set of images to model a large set of complex textures
- The proposed models are also integrated into the orginal HEVC source code
Automatic sewer network crack detection based on Deep Learning
- Collect image to create a dataset for automatic crack detection
- Perform deep learning-based automated cracks detection
- Automatically extract contextual information from frames
- Develop software that automatically generates reports based on detected cracks in sewer pipe videos.
Deep learning-based Fusarium wilt of radish detection
- Collect visible images and infrared images of radish field by drones
- Propose a fusarium wilt of radish detection system using deep learning.